For nurses with a nursing degree from outside the European Union.
The French that makes you hireable in a French hospital
You have a nursing degree. What you're missing isn't French in general — it's the French you need on a hospital shift.
That's exactly what we train: out loud, in conversation, on situations that come up on the ward every day.
We are opening access to the first people who sign up.
Does this sound familiar?
- You understand French films, but not an elderly patient explaining where it hurts.
- You know medical terms from a textbook, but at handover a colleague talks fast and uses abbreviations, and you lose the thread.
- You call the on-call doctor, and they ask you to repeat. Twice.
- You work below your qualification, because you still haven't found the nerve to go to an interview.
Your French isn't bad. You were taught the wrong French.
Shift French isn't street French.
The difference isn't vocabulary — you can learn professional terms from a list. It comes down to three things a general course never gives you:
Pace.
A handover means ten patients in fifteen minutes. There's no time to build a sentence.
Structure.
Patient information is handed over in an accepted order. A nurse who explains at length, out of order, loses her colleague's attention — and the colleague misses what matters.
Register.
You speak differently to a patient, to a doctor, to a family member, and in a written note in the chart. In a foreign language, a register mistake lands harder than you meant it to.
None of these three skills come from another hundred hours of general French. They only come from practising the situation itself.
What you will be able to do
- Give a shift handover on time, in the accepted order, without losing what matters.
- Call the on-call doctor at night and be understood the first time.
- Work out what a patient is complaining about when they use everyday words, and put it in professional terms.
- Not make a mistake on a number heard out loud — a dosage, a reading, a dosing time.
- Explain to a worried family member what can be explained, without stepping outside your scope of practice.
And know whether you are ready for the language exam — because you are assessed against the same criteria as the examiner.
The only simulator where the French of a specific hospital situation is assessed against the criteria a state examiner applies.
Real-time voice.
You speak; you don't click through answer options. The other person replies without delay — like a real person.
The job, not French in general.
Not "French for travelers" — a conversation with a patient, a call to a doctor, a shift handover.
Scored on examiner criteria.
Coherence, vocabulary range, accuracy, fluency, task achievement — the same criteria used in the state exam.
Five situations from the job
| The situation | Why it's hard | The cost of a mistake |
|---|---|---|
| Handing over a patient to a colleague, out loud | Fast pace, an accepted structure, abbreviations | A piece of information about the patient that never gets passed on |
| An urgent call to the on-call doctor | No face, no gestures, a poor line, a rushed listener, and the right degree of insistence to strike | A medical response that arrives late |
| A conversation with a patient in pain | The patient uses everyday words, sometimes unclear because of age or condition | A complaint that gets misunderstood |
| A written note in the patient's chart | A different style, accepted abbreviations, a document with legal weight | A nurse left without legal protection |
| A conversation with the patient's family | A register of courtesy and empathy, plus the ability to set a boundary | The family's trust, lost |
Numbers run through all five situations: a dosage, a blood pressure reading, a dosing time, a weight. French numbers are built in a complicated way, and heard out loud, they stay difficult even at a good level. That is why numbers are not taught as a separate lesson: they are built into every situation.
What one lesson looks like
Ten to fifteen minutes. In the messaging app where your family already texts you.
The brief
An evening shift. Six patients. You have seven minutes to hand them over to the colleague coming in for the night.
A moment from the conversation
-
You
Chambre 2, rien à signaler, elle a bien mangé ce soir. Chambre 3, la douleur reste stable, elle est sous quatorze millilitres par heure ce soir.
Room 2, nothing to report, she ate well tonight. Room 3, pain is stable, she's on fourteen millilitres an hour tonight. Two rooms in one breath, condition and events run together.
-
The colleague
Attends. Quarante, tu veux dire ? J'ai entendu quatorze.
Wait. You mean forty? I heard fourteen. The colleague cuts in and makes her repeat the number, unsure she heard it right.
-
You
Pardon, oui, quarante millilitres par heure. Je vais trop vite. Chambre 4—
Sorry, yes, forty millilitres an hour. I'm going too fast. Room 4— She corrects herself at once, then tries to move on to the next room.
-
The colleague
Vas-y, je dois y aller dans deux minutes.
Go ahead, I need to leave in two minutes. She hurries her along: room 4 never gets covered in this excerpt.
The debrief
- Completeness
- What needed to be done for the patient in room 4 was never passed on: the handover ends before it.
- Order
- She moves from room 2 straight into room 3 with no pause or transition marker: the colleague never gets a moment to confirm she's keeping up before the next patient starts.
- Number
- "Fourteen" instead of "forty" in a number. The colleague made her repeat it; in real life, she might not always get that chance.
- Time
- Nine minutes for a handover meant to take seven: by the end of the excerpt, room 4 still hasn't been covered.
This debrief isn't a separate system: it's the state examiner's grid — pronunciation included — applied to this exact situation.
Three steps
-
01
You speak out loud, in a conversation with a virtual patient, doctor or colleague.
-
02
The system hears not just the words but how you say them, and shows you exactly which sounds get in the way of being understood.
-
03
Tomorrow's lesson is built from today's mistakes: more numbers if dosages trip you up, less of what you have already mastered.
Nothing to install.
The lesson arrives on WhatsApp or Telegram, or opens in your browser. Ten to fifteen minutes before or after your shift: you do not have two free hours, and the product is built around that fact, not the other way round.
Why this didn't exist two years ago
The idea of a conversational trainer isn't new. It stayed out of reach for one reason: computers couldn't talk.
The previous generation's pipeline worked like this: a person speaks → the recording turns into text → the text gets processed → the reply is synthesised back into speech. The pause lasted three to five seconds, intonation got lost, and pronunciation mistakes disappeared as early as the first step — speech recognition is built to guess what was said despite an accent. That's not a conversation. It's an exchange of telegrams.
Models now work with voice directly.
No detour through text in between. Response delay has dropped to the level of a human reaction.
The model hears how the words are said, not just the words themselves.
Accent, intonation, hesitation. That's what makes it possible to assess pronunciation, not only whether the words were right.
French-language models have appeared in Europe.
These include, among others, work from the French company Mistral AI. Their existence frees this kind of tool from dependence on US infrastructure and allows voice to be processed within a European legal framework.
Our innovation isn't inventing these technologies — it's applying them to a problem that, until now, only a live teacher could solve.
Language, yes. Professional judgment, no.
We do not teach medicine. Our users already hold a nursing qualification and training.
The system evaluates only the language side: clarity, precision of wording, command of terminology, pronunciation. It never evaluates whether a professional decision is correct. If a user describes the wrong clinical approach in flawless French, the system scores the language and does not validate the content — that is outside its scope.
This boundary is built into the product's architecture, and it is stated clearly to the user.
Artificial intelligence doesn't operate here without human oversight
Practising nurses check every scenario
before it reaches users: does this situation really come up in practice, are these the words actually used on the ward, would a real French patient speak this way. No scenario moves forward without a professional's sign-off.
Accredited teachers calibrate the scoring.
A qualified teacher of French as a foreign language independently scores a sample of answers already scored by the system. The two sets of scores are compared: the gaps show where the model drifts from a live examiner's judgment, and become the basis for correcting it.
We are not replacing French teachers or healthcare professionals. We are creating a new kind of expert work for them: one practising nurse passes her experience, through the platform, on to far more than two or three students.
How this differs from the alternatives
| Category | What it covers | What it doesn't give you |
|---|---|---|
| Mass-market language apps | price and habit | no profession, no speaking under pressure, no scoring against an exam scale |
| AI conversation trainers | spoken practice | general language, with no French hospital context |
| Courses with a live teacher | a real professional context and accreditation | price and scheduling: dozens of two-hour sessions |
| Clinical patient simulators | training in clinical interaction | they train native speakers' judgment, not a foreign speaker's language |
| Recrute Facile | the French of a specific hospital situation, plus scoring against an exam scale, ten to fifteen minutes a day | does not teach medicine or evaluate clinical decisions; does not replace practice on the ward |
A specialised medical French course exists, and it's a good one. It also asks for dozens of hours of scheduled classes, and costs as much as a course with a teacher. A nurse working shifts can manage neither — not for lack of will, but because she has neither the hours nor the money. We work within the fifteen minutes she actually has.
Where we're heading
-
Now
Nurses: five work situations, an evaluated voice dialogue.
-
Next
Specializations: vocabulary, speech pace and typical situations differ significantly between intensive care and geriatrics, and each calls for its own scenarios and its own expert.
-
Later
Related caregiving roles, and versions of the product built for in-house training at healthcare institutions.
The technological foundation doesn't change — what changes is the content and the team of experts. The same trait makes the model applicable beyond healthcare.
Frequently asked questions
Do you replace the teacher?
No. We provide practice that price and scheduling make impossible to get with a teacher, and we bring in teachers themselves to calibrate the scoring.
What happens to my voice recordings?
Voice recordings from your practice sessions are hosted and processed in Europe, kept only as long as necessary, never shared with third parties; you can delete your history at any time.
Does this prepare you for the exam?
The evaluation applies the same criteria as a state examiner, so you can see where you stand.
Do I need to install an app?
No — the lesson arrives in a messaging app, or opens in your browser, on your phone or your computer.
My degree isn't from the EU — is this for me?
The product is built exactly for this situation.
My French isn't great — isn't it too early for me?
You can get it wrong with a virtual patient a hundred times over. That's exactly the point.
We are opening access as scenarios are ready
No scenario moves forward without sign-off from a practising nurse at a French hospital. That's slower than releasing everything at once. We think it's the right way to do it. Leave your address — we'll write to you when access opens, starting with the first people who signed up.